Magnemo vs Claude memory and Projects

Your memory lives on your computer, in plain files you can open.

IN THE TREE: test_the_network_line_passes_on_the_package_and_fails_on_a_planted_import · test_free_tier_stdlib_only

Memory saved "as a set of individual topics"; listed under Settings > Memory > Topics; "stored with encryption at rest." Each project has "its own separate memory space". [S1]

The only write verb. Lands in _staging/ as a memory candidate; never touches canon.

IN THE TREE: test_lists_exactly_four_tools · test_promotion_is_not_a_tool · test_the_author_is_the_seat_and_a_keyholder_claim_is_refused

Claude, automatically "as you chat"; user can "tell Claude to 'remember this' to save something directly." [S1]

Agent writes NEVER reach canonical stores directly — staging only, always.

IN THE TREE: test_staging_is_the_only_note_write_path · test_retrieve_excludes_staged

not stated in their docs (read 2026-10-02). Only after-save notice for sensitive topics: "a notice appears above the message box so you can review it". [S1]

Every note is markdown with provenance frontmatter (author, written, source, status, reviewed_by, supersedes, strength).

IN THE TREE: test_stage_lands_in_staging_with_provenance_and_salience · test_paperless_and_malformed_notes_refused · governance.promote

not stated in their docs (read 2026-10-02). No author, timestamp, origin or lineage fields documented.

Anything that speaks MCP over stdio mounts the same way.

IN THE TREE: test_connect_and_list_over_stdio

Text. Prompt Claude to "Write out your memories of me verbatim"; import by pasting text ("Add to memory"). "All memory data is included in data exports." [S1][S2] Export file format not stated [S3].

Python 3.11+, zero dependencies, Apache-2.0.

IN THE TREE: LICENSE

not stated in their docs (read 2026-10-02)